In this ELOQUENCE Web Café session, the discussion focused on how artificial intelligence is entering our homes through smart technologies — and what this means for privacy, fairness, multilinguality, and trust.
The session featured Jordi Luque and David Solans from Telefónica, the coordinator of the ELOQUENCE project. Together, they explored how AI is already shaping domestic environments and how future smart home technologies can become more privacy-preserving, inclusive, and trustworthy.
The conversation opened with a reflection on how AI has already become part of everyday life. Smart televisions, phones, voice assistants, speakers, gaming devices, cameras, and home automation systems are increasingly connected and intelligent. These technologies can help users control devices, manage entertainment, connect services, and automate daily routines. In that sense, AI can save time and make homes more responsive to people’s needs.
However, as David explained, the opportunities come with important risks. AI in smart homes raises several questions that are central to trustworthy AI: privacy, fairness, robustness, and explainability. If a smart device is listening, recording, or learning from user behaviour, people need to know what data is being collected, how it is used, and who can access it. This is especially sensitive in the home, where conversations are personal and may include private family, health, financial, or emotional information.
One of the key points of the discussion was data ownership. Instead of collecting all user data centrally and processing it on external servers, future systems could keep more data locally on users’ own devices. Approaches such as distributed learning and federated learning may help protect privacy by allowing models to improve without directly sharing sensitive personal data. Still, the speakers emphasised that this area remains technically complex, especially when it comes to ensuring that data can truly be deleted or “forgotten” if a user withdraws consent.
The session also addressed fairness and multilinguality. Smart home technologies often work better for high-resource languages such as English or Spanish, while users speaking lower-resource languages, regional languages, or with different accents may receive poorer performance. This can reduce trust and widen the digital gap. In multilingual households, inclusive speech technologies must be able to understand different languages, accents, and code-switching patterns, so that all users can interact with smart systems equally.
Another important theme was transparency. To build trust, companies should explain clearly what smart devices do, what data they collect, and how users can control these settings. Technical and legal language is often difficult to understand, so user-friendly communication and simple privacy controls are essential.
The discussion concluded by looking toward the future. As AI becomes more embedded in domestic life, homes themselves may eventually be designed with AI infrastructure in mind, just as they are already planned around electricity, internet, heating, and security systems. The challenge will be to ensure that these future homes are not only smart, but also safe, fair, transparent, and respectful of the people who live in them.
Listen to the full episode here.
